Parted 1.6.0-pre4

[Andrew Clausen]

Hi all,

Still haven't fixed the regression tests... plenty of other problems
to fix... :)

Changes:
* libparted/linux.c (_flush_cache): only flush partition devices if they
  aren't mounted.
* libparted/constraint.c (_constraint_get_end_range): fixed fencepost
  (out-by-one) error in maths
* libparted/disk.c (_partition_get_overlap_constraint): rewrote, this
  time running in update mode.
* (ped_disk_add_partition, ped_disk_set_partition_geom): adjusted to
  call _partition_get_overlap_constraint() in update mode.  Much
  cleaner now :)  (Also, there were subtle cases that were broken
  before, that are correct now, since neither function require
  the disk to be in update mode...)
* applied updated translations: ca da de
* libparted/disk_dos.c: applied patch from Andreas Dilger... replaced
  partition magic values with #defines
* libparted/linux.c (_disk_sync_part_table): applied patch from Andreas
  Dilger... delete all non-busy partitions and add them all back, to
  avoid overlap problems
* libparted/disk_mips.c (_handle_no_volume_header): fixes to "variable might
  be used uninitialised in this function" compile warnings.  (The wrong
  exit goto was being called from the PED_EXCEPTION_CANCEL case)
* parted.m4: don't call AC_CANONICAL_HOST, AC_REQUIRE(AC_CANONICAL_HOST)
  instead! This fixes getting an autoconf warning about AC_CANONICAL_HOST
  being invoked multiple times

Andrew